In both start() and stop(), ret is only assigned on failure. When
hypervkvpd starts or stops successfully, return "$ret" expands to an
empty string and causes:
/etc/init.d/S10hyperv: return: line 31: Illegal number:
Those double quotes were added in Buildroot commit [1], to fix a
new ShellCheck warning at that time. This was not a complete fix.
Only removing the double quote would reintroduce the ShellCheck
warning. This would also reintroduce a check-package error.
Since a bare return is equivalent to a "return 0", this commit
also initializes with ret=0. Doing so will tell ShellCheck "ret" is
an integer. Therefore, the ShellCheck warning will no longer be
reported.
This commit fixes the invalid return value by removing the double
quotes and initialzing "ret=0".
[1] c4173d8b08
Signed-off-by: Benjamin DeCamp <benjamin8532@protonmail.com>
[Julien:
- add "ret=0" initialization in script to fix check-package error
- add extra info in the commit log
]
Signed-off-by: Julien Olivain <ju.o@free.fr>
enscript currently fails to build with musl with gcc >= 15. In order
to fix this, we need to bring a number of patches from upstream, and
add 2 others that were submitted upstream.
From upstream, we bring
0002-Add-CFLAG-std-c89-so-it-compiles-with-the-old-standa.patch, which
switches to -std=c89 to get the compiler back to "old" behavior.
However, as this commit patches configure.ac, we need to autoreconf,
but autoreconf is broken, so we also take
0003-Automake-1.12-and-up-no-longer-supports-pre-ANSI.patch from
upstream, which drops a problematic autoconf macro.
However, once you drop this problematic autoconf macro, the PROTOTYPES
define is never set by anything, causing the __P macro to no longer be
defined properly. This is fixed by
0004-Fix-prototype-detection-when-__STDC__-is-defined-but.patch that
we have submitted upstream.
Once you're there, you realize that switching to -std=c89 has the side
effect that musl's <limits.h> no longer defines PATH_MAX, because it
needs one of:
#if defined(_POSIX_SOURCE) || defined(_POSIX_C_SOURCE) \
|| defined(_XOPEN_SOURCE) || defined(_GNU_SOURCE) || defined(_BSD_SOURCE)
and a side effect of -std=c89 is that none of these is defined
anymore. So we introduce 0005-Use-std-gnu89-instead-of-std-c89.patch,
which switches to -std=gnu89. This patch has also been submitted
upstream.
With all of these efforts, we get a successful build on musl with gcc
>= 15.
This commit needs to be backported to Buildroot versions that support
gcc 15.x, so that means the currently maintained 2026.x branches, but
not 2025.02 as only up to gcc 14.x was supported then.
Fixes:
https://autobuild.buildroot.org/results/d39d14bbbb3a51d67fe962b877c7f66ff1204ecf/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Backport the fix for CVE-2026-66035.
The ETM decrypt path does not validate the received packet length before
calculating the decrypt buffer size. A malformed packet can therefore
lead to a heap overflow.
Use Debian's libssh2 1.11.1 backport of the upstream fix.
Signed-off-by: Stefan Müller <stefan.mueller@rey-technology.com>
[Julien: add links to Debian patches]
Signed-off-by: Julien Olivain <ju.o@free.fr>
Backport the fix for CVE-2026-66034.
The publickey subsystem does not sufficiently validate the length of a
server-controlled comment field. A malformed response can therefore
cause an out-of-bounds read.
Use Debian's libssh2 1.11.1 backport of the upstream fix.
Signed-off-by: Stefan Müller <stefan.mueller@rey-technology.com>
[Julien: add links to Debian patches]
Signed-off-by: Julien Olivain <ju.o@free.fr>
Backport the fix for CVE-2026-66033.
The OpenSSL AES-GCM cipher path lacks runtime bounds checks around the
input block size. A malformed packet can therefore lead to an
out-of-bounds read or write.
Use Debian's libssh2 1.11.1 backport of the upstream fix.
Signed-off-by: Stefan Müller <stefan.mueller@rey-technology.com>
[Julien: add links to Debian patches]
Signed-off-by: Julien Olivain <ju.o@free.fr>
Backport the fix for CVE-2026-66032.
A SFTP error path can leave a dangling pointer after freeing the
response buffer, which may result in a double free on subsequent error
handling.
Use Debian's libssh2 1.11.1 backport of the upstream fix.
Signed-off-by: Stefan Müller <stefan.mueller@rey-technology.com>
[Julien: add links to Debian patches]
Signed-off-by: Julien Olivain <ju.o@free.fr>
Backport the SFTP symlink bounds checking fix for CVE-2025-15661.
The initial fix requires the LIBSSH2_UNCONST compatibility backport on
libssh2 1.11.1. Also include the upstream follow-up fixing
SSH_FXP_STATUS handling introduced by the initial security fix.
The patches are based on the upstream fixes and Debian's libssh2 1.11.1
backports.
Signed-off-by: Stefan Müller <stefan.mueller@rey-technology.com>
[Julien: add links to Debian patches]
Signed-off-by: Julien Olivain <ju.o@free.fr>
Since upstream commit 89a86dcb0a3248606824de50f5c63f61cfe0369c (first
release: 106) if cargo exists on PATH the Dracut configure script
enables building dracut-cpio by default, and calls "cargo --version"
to check if cargo works. This fails on the autobuilders:
error: rustup could not choose a version of cargo to run, because one wasn't specified explicitly, and no default is configured.
help: run 'rustup default stable' to download the latest stable release of Rust and set it as your default toolchain.
dracut couldn't find cargo for dracut-cpio build
The affected configs either don't have BR2_PACKAGE_HOST_RUSTC enabled,
or build-time.log.gz shows host-rustc was not installed before the
host-dracut build, so presumably the "cargo" that produces the rustup
error is an external one already installed on the autobuilders.
To fix this, enable dracut-cpio only if BR2_PACKAGE_HOST_RUSTC=y, and
add a dependency on host-rustc in that case. According to the
documentation [1, see "enhanced_cpio"] dracut-cpio is supposed to
optimize archive creation for copy-on-write filesystems, so it should
not matter much for Buildroot. The --disable-dracut-cpio option was
added in upstream commit 4a4ab928a49e81e02104ec5466160664e59c3965
(same release).
Fixes: https://autobuild.buildroot.org/results/5f557d708cce997e7f039f17e30640b02ba9180a/
Fixes: https://autobuild.buildroot.org/results/f04ca3c4598f62a7e87d84bc111eb8b161b34a70/
(and more)
[1] https://dracut-ng.github.io/dracut/man/dracut.conf.5.html#_configuration_options
Signed-off-by: Fiona Klute (Othermo GmbH) <fiona.klute@gmx.de>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Backport the upstream fix for a heap buffer overflow in
convert_fname() when growing the iconv output buffer.
Backport to: 2025.02.x
Signed-off-by: Stefan Müller <stemu86@gmx.ch>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Backport the upstream fix for integer overflows while parsing
Content-Range headers, together with the follow-up fix using
strtoll() for wgint values.
Backport to: 2025.02.x
Signed-off-by: Stefan Müller <stemu86@gmx.ch>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Backport the upstream fix for a buffer underflow in
clean_metalink_string(), together with the two required follow-up
fixes for the inverted whitespace check and missing ctype.h include.
Backport to: 2025.02.x
Signed-off-by: Stefan Müller <stemu86@gmx.ch>
Signed-off-by: Julien Olivain <ju.o@free.fr>
The xilpm_runtime_lib is not enabled by default in the versal2_plm Makefile:
97f2baf7f6/lib/sw_apps/versal_plm/src/versal_2ve_2vm/Makefile (L13)
Without it, there is a silent runtime failure.
Add config XILPM_RUNTIME_LIB=SUBSYS to make sure the xilpm_runtime_lib is
correctly configured and included to fix the problem.
Signed-off-by: Neal Frager <neal.frager@amd.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
distribution-registry calls pthread_getattr_np() which is only available
with NPTL; i.e. always available with glibc (where it originates from,
since 2.2.3), always available with musl (which has had it since 0.9.10
in 2013), and only available when uClibc has NPTL (since 1.0.0 in 2015).
Fixes: https://autobuild.buildroot.org/results/9395500a8baee6c6142f96d7bc97e81725c2e754/
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
The Arc specific gdb version was removed by commit [1]
but we still have the TestGdbArc that was testing this
version of gdb.
We can now safely remove TestGdbArc.
[1] 0b3d526226
Signed-off-by: Romain Naour <romain.naour@smile.fr>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Some examples embed CSP views, whose C++ sources are generated at build
time by drogon_ctl. When cross-compiling, CMake looks the tool up in
PATH, so the build fails with:
[ 77%] Generating HelloView.h, HelloView.cc
/bin/sh: 1: drogon_ctl: not found
make[3]: *** [examples/CMakeFiles/helloworld.dir/build.make:74: examples/HelloView.h] Error 127
Add host-drogon to the dependencies, as it installs drogon_ctl in
$(HOST_DIR)/bin.
Fixes:
- https://autobuild.buildroot.org/results/b8f38b0645932cb5506515d6d313b64d824c8ce0
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Build errors were found by the Gitlab pipelines with these defconfigs:
- bootlin-m68k-68040-uclibc
CMake Error at cmake/scripts/linux/ArchSetup.cmake:50 (message):
Unknown CPU: m68k
- bootlin-s390x-z13-glibc
CMake Error at cmake/scripts/linux/ArchSetup.cmake:50 (message):
Unknown CPU: s390x
Backport an upstream commit from the upcoming Piers branch to fix the
restriction in ArchSetup.cmake.
This caused a different build error on m68k later on:
/builds/bkuhls/buildroot/br-test-pkg/bootlin-m68k-68040-uclibc/build/kodi-21.3-Omega/xbmc/utils/MathUtils.h:142:5:
error: unknown register name ‘st’ in ‘asm’
142 | __asm__ __volatile__ (
because m68k is not part of the list of archs to disable asm code:
https://github.com/xbmc/xbmc/blob/Omega/xbmc/utils/MathUtils.h#L26
Upstream rejected to add m68k there:
https://github.com/xbmc/xbmc/pull/22519https://github.com/xbmc/xbmc/pull/22357#issuecomment-1368358471
so we disable m68k in BR2_PACKAGE_KODI_ARCH_SUPPORTS.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
When a package defines $(PKG)_FLAT_STACKSIZE, ELF2FLT_FLAGS contains
-Wl,-elf2flt="-r -s<stack-size>". The embedded quotes are needed to
keep both elf2flt options in single linker argument.
However, many package Makefiles wrap $(TARGET_CFLAGS) in double quotes,
for example:
CFLAGS="$(TARGET_CFLAGS)"
After expansion, the embedded quote terminates the outer CFLAGS quote.
As a result, the shell interprets "-s<stack-size> ..." as a command
instead of passing it to the compiler.
Pass -r and -s<stack-size> in separate -Wl arguments instead. This
avoids embedded quotes; GCC forwards both -elf2flt options to
ld-elf2flt, which collects them before invoking elf2flt.
This got broken by commit
04d7ea4720 ("package: Makefile.in: fix
elf2flt invocation options"), which by adding -r as an elf2flt
argument, did not correctly handle -s$($(PKG)_FLAT_STACKSIZE).
Signed-off-by: Fengwei Tan <tfx2001@outlook.com>
[Thomas: improve commit message]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
wine.mk passes --with-wayland whenever BR2_PACKAGE_WAYLAND is enabled,
but nothing guarantees the rest of what wine's Wayland test needs is in
the configuration. That test is:
WINE_NOTICE_WITH(wayland, [test -z "$WAYLAND_CLIENT_LIBS" \
-o -z "$WAYLAND_SCANNER" -o -z "$XKBCOMMON_LIBS" \
-o -z "$XKBREGISTRY_LIBS" -o "$ac_cv_header_linux_input_h" = "no"], ...)
and because --with-wayland is passed explicitly, WINE_NOTICE_WITH turns
into AC_MSG_ERROR rather than a notice.
So wine needs libxkbcommon, and it needs the libxkbregistry part of it,
which is only built when libxml2 is available. Select both when Wayland
support is enabled, and add libxkbcommon to the build dependencies.
Note that libxml2 is not a direct dependency of wine, it only has to be
in the configuration so that libxkbcommon builds libxkbregistry; the
build ordering is handled by libxkbcommon's own dependency on libxml2.
Signed-off-by: Alsey Coleman Miller <alseycmiller@gmail.com>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
libxkbregistry is the keyboard layout catalogue half of the library. It
parses the XML layout registry and so needs libxml2, which is presumably
why it was disabled unconditionally rather than wired to a dependency.
wine needs it. Its configure.ac requires XKBREGISTRY_LIBS alongside
wayland-client, wayland-scanner, xkbcommon and linux/input.h before it
will build the Wayland driver, and wine.mk passes --with-wayland for any
build with BR2_PACKAGE_WAYLAND - which turns that notice into a hard
error:
checking for wayland-client.h... yes
checking for wl_display_connect in -lwayland-client... yes
checking for wayland-scanner... .../host/bin/wayland-scanner
checking for xkb_context_new in -lxkbcommon... yes
checking for wayland-egl.h... yes
checking for wl_egl_window_create in -lwayland-egl... yes
configure: error: Wayland development files not found, the Wayland
driver won't be supported.
This is an error since --with-wayland was requested.
Every other term of that test passes; only XKBREGISTRY_LIBS is empty, so
wine and wayland together could not be built on any architecture.
Gated on BR2_PACKAGE_LIBXML2 rather than turned on outright, because
meson.build takes dependency('libxml-2.0') unconditionally once
enable-xkbregistry is set, so a target without libxml2 would fail to
configure.
Regarding since when this is broken, three pieces had to come together:
- libxkbcommon has passed -Denable-xkbregistry=false since commit
1791bc30a5 ("package/libxkbcommon: bump version to 1.0.1", Sep 2020),
i.e. Buildroot 2020.11. libxkbregistry has therefore never been built
in Buildroot.
- wine's configure gained the XKBREGISTRY_LIBS term in its Wayland
test in wine 9.0, with upstream commit d64ea8e4a6c9
("winewayland.drv: Enumerate Xkb layouts and create matching HKL.",
Nov 2023).
- wine.mk started passing --with-wayland in commit 7cb49e7712
("package/wine: bump to version 9.19", Oct 2024), which is what turns
the missing XKBREGISTRY_LIBS from a notice into a hard error.
The breakage therefore dates from Buildroot 2024.11, and every branch
since is affected, including the LTS one: 2025.02.x carries wine 10.0,
whose configure has the XKBREGISTRY_LIBS check, together with
libxkbcommon 1.9.2 built with -Denable-xkbregistry=false, and its wine.mk
passes --with-wayland. 2025.05.x and 2025.08.x are in the same state.
A backport to 2025.02.x is thus needed.
Signed-off-by: Alsey Coleman Miller <alseycmiller@gmail.com>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Fixes the following security issues:
- x/mod/sumdb/tlog: fix transparency log tile verification bypass
A malicious GOPROXY was previously capable of forging up to two sumdb
tiles that allow for a requested module to bypass the GOSUMDB check and
persist attacker-controlled module content to a local Go module cache.
This attack allows for a malicious GOPROXY to serve malicious module
content that cannot be detected by evaluating the transparency log.
All tiles are now correctly verified against their parents.
In order to determine if you have been affected:
rm -r go.sum go.work.sum vendor/ && go mod tidy
Thanks to Filippo Valsorda (Geomys) for reporting this issue.
This is CVE-2026-56865 and Go issue https://go.dev/issue/80744.
- x/mod/sumdb: ignore unrelated, unauthenticated hashes in Lookup
A malicious GOSUMDB was capable of serving arbitrary module content not
contained within the transparency log.
This attack allows for a coordinating GOPROXY and GOSUMDB to serve a
client malicious module content that cannot be detected by evaluating
the transparency log.
In order to determine if you have been affected:
rm -r go.sum go.work.sum vendor/ && go mod tidy
Thanks to mundur for reporting this issue.
This is CVE-2026-56864 and Go issue https://go.dev/issue/80745.
- encoding/xml: add recursion depth guard during decode
Previously, DecodeElement would reset the depth counter causing it to
never fire; this could lead to stack exhaustion.
This is CVE-2026-56859 and Go issue https://go.dev/issue/80481.
- net/http: apply ReadHeaderTimeout when doing unencrypted HTTP/2 check
When a server is configured to support unencrypted HTTP/2, it reads a few
bytes from each new connection to see if they contain the HTTP/2 client
preface. Previously, this was being done with no timeout applied.
ReadHeaderTimeout is now applied for this.
This is CVE-2026-56853 and Go issue https://go.dev/issue/80205.
- net/url: avoid quadratic complexity in resolvePath
Previously, resolving relative paths containing parent directory (..)
segments performed string conversions and buffer rewrites on each step,
resulting in quadratic time complexity and high memory allocation
overhead.
Now, path resolution operates on a byte buffer using index-based
backtracking for .. segments, eliminating the quadratic time complexity
and significantly reducing memory allocations.
This is CVE-2026-56860 and Go issue https://go.dev/issue/80494.
- golang.org/x/net/dns/dnsmessage: panic when parsing invalid SVCB record
Parsing an invalid SVCB or HTTPS RR can panic when the size of a
parameter value overflows the message buffer.
Thanks to Mundur (https://github.com/M0nd0R) for reporting this issue.
This is CVE-2026-46600 and Go issue https://go.dev/issue/79795.
- crypto/tls: limit handshake messages we are willing to accept post-handshake
Previously, we always counted handshake messages, such as KeyUpdate, as
state-advancing, regardless of whether a handshake has been completed or
not. As a result, a malicious client can keep sending KeyUpdate messages
to force the server to keep performing key derivation operations
indefinitely.
Thanks to Qi Deng of Aurascape.ai for reporting this issue.
This is CVE-2026-56862 and Go issue https://go.dev/issue/80528.
- html/template: fix Javascript regexp context tracking
Previously, pathological inputs could close an unescaped / early,
allowing for attack-controlled data to inject arbitrary content,
potentially leading to XSS.
Thanks to Ali Sherif for reporting this issue.
This is CVE-2026-56858 and Go issue https://go.dev/issue/80435.
- x/net/idna: failure to reject ASCII-only Punycode-encoded labels
The ToASCII and ToUnicode functions incorrectly accepted Punycode-encoded
labels that decode to an ASCII-only label. For example,
ToUnicode("xn--example-.com") incorrectly returned the name "example.com"
rather than an error.
The idna package implements the processing algorithm from UTS 46. Older
versions of UTS 46 included a specification bug which permitted multiple
ASCII labels to decode to the same Unicode label. UTS 46 revision 33
fixed the specification bug. The idna package now implements the updated
specification.
This behavior can lead to privilege escalation in programs using the idna
package. For example, a program which performs privilege checks on the
ASCII hostname may reject "example.com" but permit "xn--example-.com".
If that program subsequently converts the ASCII hostname to Unicode, it
will inadvertently permits access to the Unicode name "example.com".
Thanks to KC1zs4 (https://github.com/KC1zs4) for reporting this issue.
This is CVE-2026-39821 and Go issue https://go.dev/issue/78760.
- encoding/asn1: enforce maximum recursion depth
Enforce a recursion limit in Unmarshal to prevent stack exhaustion when
parsing deeply-nested, recursive structures.
Thanks to Marwan Atia (marwansamir688@gmail.com) for reporting this issue.
This is CVE-2026-33818 and Go issue https://go.dev/issue/80405.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Buildroot commit 0f2e9cc878 bumped the
package from 3.11.0 to 3.12.0. This bump includes upstream commit
9f8945251d
which added the usage of enums not present in older kernel versions.
The Gitlab pipelines caught the build errors with the defconfig
bootlin-aarch64-glibc-old:
lib/route/nh_encap_ila.c:50:19: error: ‘ILA_ATTR_IDENT_TYPE’ undeclared
(first use in this function)
lib/route/nh_encap_ila.c:53:19: error: ‘ILA_ATTR_HOOK_TYPE’ undeclared
(first use in this function)
Both enums were added to the Linux kernel in version 4.15:
fddb231ebe70d5aef48a
Add upstream commit to fix the problem.
Signed-off-by: Bernd Kuhls <bernd@kuhls.net>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Bugfix release with large number of (security) fixes.
HAProxy 2.6.32 was released on 2026/07/29. It added 33 new commits
after version 2.6.31.
As for the 2.8.27, the announce is an expurgated copy-paste of the 3.4.3
announce:
* stats: Two issues about the stats page, reported by Red Hat/AISLE
Research, were fixed.
Proxies updated through the stats page while in "stats admin" mode were
not subject to the "stats scope" filtering, meaning a scope meant to
restrict which proxies are visible/actionable could be silently bypassed
on POST requests.
Separately, POST requests to the stats interface did not validate that the
Origin (or Referer) header matched the Host, which is now checked to
mitigate CSRF attacks.
* ssl-gencert: A memory leak on every certificate generation was fixed.
Two temporary buffers were not freed after generating a certificate on the
fly, leaking memory each time a new SNI triggered certificate
generation. This issue was reported by Red Hat/AISLE Research.
* sample/protobuf: buffer overflows after pointer-shift converters, reported
by Red Hat/AISLE Research and Charles Vosburgh, were fixed.
Several converters (protobuf/ungrpc field extraction, ltrim())
move the sample's data pointer forward on success but did not shrink the
sample's recorded buffer capacity accordingly. A converter chained
afterwards that relies on that capacity (e.g. padding via memset()) could
then write past the end of the buffer, leading to heap corruption or a
worker crash. All the affected converters now adjust the capacity
together with the pointer.
* protobuf: A nested-path validation bypass reported by Red Hat/AISLE
Research was fixed.
The protobuf field lookup used for the protobuf()/ungrpc() converters did
not strictly enforce hierarchical boundaries, so a flat sibling field
could incorrectly satisfy a nested-path lookup (e.g. matching a root-level
field as if it were nested under a parent). The lookup was rewritten as a
strict, non-recursive path walker that correctly bounds each nesting
level.
Separately, a crash because of deprecated protobuf group wire types was
fixed. These wire types are now explicitly rejected.
* http-fetch: Two crashes reachable from health-check configurations were
fixed.
"res.body"/"res.hdr"/... and similar response fetches assumed the
health-check receive buffer always held an HTX message, which is only true
for actual HTTP checks; on a plain TCP check, a hostile/misbehaving server
could craft the first bytes of its reply to be misinterpreted as HTX
internal fields, causing a wild read and worker crash (or leaking
arbitrary process memory).
Separately, "capture.req.hdr"/"capture.res.hdr" only validated the upper
bound of their index argument, so a negative capture id was accepted at
boot and dereferenced an out-of-bounds array entry at runtime, crashing
the worker on the very first request.
* slz: Several issues were fixed in the SLZ library.
A stream alternating many literals in the 144-255 range with cheap
back-references could keep inflating indefinitely instead of falling
back to a stored block, exceeding the library's documented worst-case
output size by several percent. A new accounting mechanism now bounds
this overhead. Practical impact on haproxy requires tune.bufsize above
~43 kB with the default reserve.
Five small correctness fixes inherited from upstream libslz were also
backported: Avoid reading up to a few bytes past the end of very short
inputs on architectures without fast unaligned access; stop appending an
extra, misplaced block to an already-finished deflate/gzip/zlib stream
(which could corrupt the trailing checksum in ~2% of fuzzed streams); fix
the Adler32 checksum accumulator sign handling on 32-bit systems
(affecting the zlib format only); avoid an undefined-behaviour signed left
shift when assembling input words byte by byte; and use the exact bit cost
when deciding whether to emit the last literals of a block as a stored
block, avoiding compressed output slightly larger than the documented
worst case.
* peers: A heap overflow when replicating large stick-table dictionary
entries was fixed.
peer_prepare_updatemsg() never verified that a stick-table entry's
dictionary value (e.g. server_key, up to ~16 kB) actually fit in the
update message being built. Since the peers protocol is plain-text and
unauthenticated, a rogue or compromised peer could plant an oversized
entry that overflows the 16 kB trash buffer as soon as the victim
replicates ("teaches") it, confirmed as a heap-buffer-overflow write. The
function now checks the available room before encoding and fails cleanly
if it doesn't fit. This was reported and fixes by Matt Suiche from Tolmo
Inc.
And, as usual, the bunch of minor fixes here and there, mainly raised during
AI-assisted code reviews. Most were never noticed:
* HTX API: Some bugs about how the HTX API was used were fixed here and
there.
* http-act: Double-frees and a couple of state bugs on parsing errors were
fixed.
* http-fetch/http-ana/http-htx: Few out-of-bounds reads were fixed.
* http-conv: The last input character could be lost when calling url-dec
converter, when the input buffer was full. This was fixed by failing the
converter in that case.
* mux-h1: An extra 200ms delay was observed on some H2-to-H1 messages
because the end of the message was not always properly detected. This
case is now properly handled.
* sample: An edge case in be2hex() was fixed.
For more details, see the announcement:
https://www.mail-archive.com/haproxy@formilux.org/msg47353.html
Signed-off-by: Fred Lefranc <fred.lefranc.evs@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
